fix: support more scenarios for extract_domain

Author: hatamiarash7

src/functions/extract_domain.cpp

@@ -41,14 +41,14 @@ namespace duckdb
             Connection con (db);
 
             // Extract the host from the URL
-            std::regex host_regex (R"(^(?:(?:https?|ftp|rsync):\/\/)?([^\/\?:]+))");
+            std::regex host_regex (R"(^(?:(?:https?|ftp|rsync):\/\/|mailto:)?((?:[^\/\?:#@]+@)?([^\/\?:#]+)))");
             std::smatch host_match;
             if (!std::regex_search (input, host_match, host_regex))
             {
                 return "";
             }
 
-            auto host = host_match[1].str ();
+            auto host = host_match[host_match.size () - 1].str ();
 
             // Split the host into parts
             std::vector<std::string> parts;

test/sql/extract_domain.test

@@ -194,3 +194,86 @@ example.com.ac
 example.com
 example.com
 example.com.ac
+
+# Test IP addresses
+query I
+SELECT extract_domain('http://192.168.1.1');
+----
+(empty)
+
+query I
+SELECT extract_domain('http://192.168.1.1:8080');
+----
+(empty)
+
+query I
+SELECT extract_domain('192.168.1.1');
+----
+(empty)
+
+# Test IPv6 addresses
+query I
+SELECT extract_domain('http://[2001:0db8:85a3:0000:0000:8a2e:0370:7334]');
+----
+(empty)
+
+query I
+SELECT extract_domain('[2001:0db8:85a3:0000:0000:8a2e:0370:7334]');
+----
+(empty)
+
+# Test URLs with query parameters
+query I
+SELECT extract_domain('https://example.com?param=value');
+----
+example.com
+
+query I
+SELECT extract_domain('https://example.com/path?param=value');
+----
+example.com
+
+# Test URLs with fragments
+query I
+SELECT extract_domain('https://example.com#section');
+----
+example.com
+
+query I
+SELECT extract_domain('https://example.com/path#section');
+----
+example.com
+
+# Test combined query parameters and fragments
+query I
+SELECT extract_domain('https://example.com?param=value#section');
+----
+example.com
+
+# Test special and edge cases
+query I
+SELECT extract_domain('');
+----
+(empty)
+
+query I
+SELECT extract_domain(NULL);
+----
+NULL
+
+query I
+SELECT extract_domain('localhost');
+----
+(empty)
+
+# Test scheme with no authority
+query I
+SELECT extract_domain('file:///path/to/file');
+----
+(empty)
+
+# Test mailto URLs
+query I
+SELECT extract_domain('mailto:[email protected]');
+----
+example.com